4:37 AM
7 hours later…
11:14 AM
@user64494 Before, I tried to convince you that `DiracDelta[x] DiracDelta[y]` is meaningful by referring to Fubini's theorem (without mentioning Fubini's name).
Maybe you find the following construction more convincing:
Maybe you find the following construction more convincing:
The tensor product $\mathcal{D}(\mathbb{R}) \otimes \mathcal{D}(\mathbb{R})$ is dense in $\mathcal{D}(\mathbb{R}^2)$, hence we may write $\mathcal{D}(\mathbb{R}^2) = \mathcal{D}(\mathbb{R}) \overline{\otimes} \mathcal{D}(\mathbb{R})$.
We can define $\delta \otimes \delta \in (\mathcal{D}(\mathbb{R}) \otimes \mathcal{D}(\mathbb{R}))'$ on simple vectors $\varphi \otimes \psi$ by $\langle \delta \otimes \delta, \varphi \otimes \psi\rangle := \langle \delta ,\varphi \rangle \cdot \langle \delta , \psi \rangle$. Extending it continuously leads to a distribution $\delta \otimes \delta \in \mathcal{D}'(\mathbb{R})$.
When inserting arguments, we obtain $(\varphi \otimes \psi)(x,y) = \varphi(x) \, \psi(y)$. In order to write that into an integral over $\mathbb{R}^2$, one has to indicate which $\delta$ is in which variable.
So the notation
$$
\iint_{\mathbb{R}^2} \delta(x) \, \delta(y) \, \varphi(x) \, \psi(y) \, \mathrm{d} x\, \mathrm{d} y = \langle \delta \otimes \delta, \varphi \otimes \psi\rangle
$$
or more general
$$
\iint_{\mathbb{R}^2} \delta(x) \, \delta(y) \, f(x,y)\, \mathrm{d} x\, \mathrm{d} y = \langle \delta \otimes \delta, f\rangle,
$$
while not meeting your high standards of aesthetics, can be still interpreted in a meaningfull way. Since Mathematica has currently no other way of formalizing the pairing $\rangle \cdot, \cdot \langle$ between distributions and test functions, it simply has to use
$$
\iint_{\mathbb{R}^2} \delta(x) \, \delta(y) \, \varphi(x) \, \psi(y) \, \mathrm{d} x\, \mathrm{d} y = \langle \delta \otimes \delta, \varphi \otimes \psi\rangle
$$
or more general
$$
\iint_{\mathbb{R}^2} \delta(x) \, \delta(y) \, f(x,y)\, \mathrm{d} x\, \mathrm{d} y = \langle \delta \otimes \delta, f\rangle,
$$
while not meeting your high standards of aesthetics, can be still interpreted in a meaningfull way. Since Mathematica has currently no other way of formalizing the pairing $\rangle \cdot, \cdot \langle$ between distributions and test functions, it simply has to use
4 hours later…
2:59 PM
@HenrikSchumacher You are wasting your time: user64494 has a history of commenting on every post that mentions distributions (check their comment history). They always claim that the OP misunderstands distributions which, unfortunately, is never the case: it is user64494 who does not understand distributions.
2
this behaviour is not only very annoying and condescending, but also somewhat dangerous: people unfamiliar with user64494's behaviour might think that he/she is right, and OP is wrong.
(and there is always a downvote from user64494 on correct/good posts, which is also problematic to some extent)
« first day last day (15 days later) »
Transcript for
Sep26
Sep '1827
Oct12
Discussion between Henrik Schumacher …
Imported from a comment discussion on mathematica.stackexchang...